method of automatic diagnosis of patient condition and prediction of outcome after cochlear implantation

摘要

FIELD: medicine.;SUBSTANCE: method includes state testing by providing audio and voice signals and analyzing the responses, and requires at least two such surveys at different times. The resulting value of each of the tests performed in one survey is assigned to the coordinate of the point of the n-dimensional space, which is correlated with the current state of the patient. At that, n=k*N, where k is the number of tests performed in one survey, N is the survey sequence number. After that, a self-organizing neural network is constructed from the mentioned points of the n-dimensional space. The points are separated based on the minimized difference between the neuron input vector and its weight by a previously unknown number of classes corresponding to the patient's state. Based on the patient's migration of between classes in repeated surveys, the effectiveness of ongoing rehabilitation activities is determined.;EFFECT: invention allows to increase the accuracy of diagnosis of the patient's current state, increase the likelihood of a reliable prognosis of the rehabilitation process and its dynamics, extend the time interval for prediction of the long-term outcomes of cochlear implantation.;5 cl, 3 ex, 3 tbl, 7 dwg